The optimal immunosuppressive regimen for recipients of expanded criteria donor (ECD) kidneys has not been identified. In this single-center study, 171 recipients of ECD kidney transplants were randomized to receive antithymocyte globulin induction, and delayed introduction of reduced dose tacrolimus, prednisone and everolimus (r-ATG/EVR, n = 88), or mycophenolate (r-ATG/MPS, n = 83). No cytomegalovirus (CMV) pharmacological prophylaxis was used. The primary endpoint was the incidence of CMV infection/disease at 12 months. Secondary endpoints included treatment failure [first biopsy-proven acute rejection (BPAR), graft loss, or death] and safety. Patients treated with EVR showed a 89% risk reduction (13.6 vs. 71.6%; HR 0.11, 95% CI 0.06-0.220, P < 0.001) in the incidence of first CMV infection/disease. Incidences of BPAR (16% vs. 5%, P = 0.021), graft loss (11% vs. 1%, P = 0.008), death (10% vs. 1%, P = 0.013), and treatment discontinuation (40% vs. 28%, P = 0.12) were higher in the r-ATG/EVR, leading to premature study termination. Mean glomerular filtration rate was lower in r-ATG/EVR (31.8 ± 18.8 vs. 42.6 ± 14.9, P < 0.001). In recipients of ECD kidney transplants receiving no CMV pharmacological prophylaxis, the use of everolimus was associated with higher treatment failure compared with mycophenolate despite the significant reduction in the incidence of CMV infection/disease (ClinicalTrials.gov.NCT01895049).